Landsat images assist in mapping Caspian bathymetry
Oryx Energy turned to satellite imagery to gauge water depths along the shallow coast of the Caspian Sea in the Republic of Kazakhstan where water level changes have rendered printed bathymetric maps obsolete. "The northern Caspian Sea is one of the world´s most prospective areas for oil exploration, but wind-driven fluctuations in water depth can exceed one meter and make movement of seismic and drilling equipment a formidable challenge," said Kurt Reisser, chief geologist for
Shallow depths, wind-driven tides create problems for seismic vessels
Kevin P. CorbleyOryx Energy turned to satellite imagery to gauge water depths along the shallow coast of the Caspian Sea in the Republic of Kazakhstan where water level changes have rendered printed bathymetric maps obsolete.
Contributing Editor, is a Denver, Colorado (US) -based information specialist specializing in remote sensing, geographic information systems and GPS (Tel: 303-987-3979; email:KCorbley@aol.com).
"The northern Caspian Sea is one of the world´s most prospective areas for oil exploration, but wind-driven fluctuations in water depth can exceed one meter and make movement of seismic and drilling equipment a formidable challenge," said Kurt Reisser, chief geologist for Oryx Kazakhstan.
Oryx and its partner, Exxon, asked Falcon Information Technologies of Dallas to apply satellite image processing technology to create up-to-date bathymetric maps of its Mertvyi Kultuk exploration block, which is located about 40 kilometers south of the giant Tengiz Field. Falcon had worked previously with Oryx making geologic maps from Landsat Thematic Mapper (TM) images for international exploration projects.
The bathymetric information had to be accurate in order to plan an extensive seismic survey in the Bay of Komsomolets, and to evaluate transportation options for future drilling in the block. For the maps to be useful in the shallow project area where depths average less than 10 meters, Oryx determined depth accuracy had to be within one meter. After searching literature on satellite image processing in marine environments, Falcon was not optimistic about the mapping project.
"We were not sure satellite images could map water depths that accurately," said Falcon president Warren Thetford. "It had been done in the clear waters of the Caribbean but not in turbid waters like the Caspian Sea."
New techniqueBathymetric maps can be made from multispectral TM images with digital image processing systems by correlating known water depth measurements at a limited number of ground control points with variations in the spectral signature of the image. Results of the correlation can be extrapolated quite accurately to estimate depths throughout the image.
Previous image-derived water depth maps had been made based on the concept that the spectral variations used to infer relative depths were created only by light interaction with the seafloor. Thetford and his chief geologist, Francis Buckmeier, saw problems in this theory.
"In clear water, light bouncing off the bottom may provide a spectral signature that varies with depth," said Thetford, "but in turbid water, suspended sediment and algae may obscure the bottom and affect spectral response, resulting in incorrect relative depth estimates."
Falcon technicians reasoned that by applying some slightly more complex, yet commonly used, image processing tech niques, they could minimize effects of sediment and algae. They used TNT mips, a commercial image processing package from MicroImages, of Lincoln, Nebraska (US), to process the TM scenes. This software was chosen because it is known for its versatility in handling multiple types of digital data in virtually any format.
The ability to process diverse data sets was critical to the project´s success because the actual depth measurements had been taken with GPS at control points during seismic planning and were stored as point files in CAD format. Later in the process, these would have to be correlated digitally with the raster format Landsat image files to derive the extrapolated depth map.
Processing the imagesThe Landsat TM sensor measures reflected energy in six wavelengths, each sensitive to many specific ground and water conditions. Its visible and near infrared bands (bands 1-4) respond to algae and suspended sediment in water. The visible blue wavelength (band 1) is particularly effective at penetrating water and reflecting off the bottom.
"We applied basic band ratio and principal component analysis routines to these bands in the image processing software," said Buckmeier. "These separated out the spectral response of the algae and sediment and helped us determine which spectral signatures really related to light reflectance off the bottom."
The output from these computerized analysis routines was a file of ratio variables. There were eight variables for each pixel in the image. The technicians then imported the point file of known GPS depth measurements into TNTmips and merged it with the ratio variables so that each control point corresponded to a set of ratio variables. In the initial pilot project covering a small area, there were 150 GPS depth measurements.
Falcon applied a standard correlation routine in a statistics software package to the depth values and ratio variables. This routine determined which variables corresponded most accurately with depth. Based on this information, the technicians wrote a simple equation in the image processing software that used the known depth values and the selected variables to extrapolate bathymetric measurements for every pixel in the image. They created a color image map of the Oryx project area and shaded the colors by various depth increments.
"The water depth values were accurate to within one third of a meter," said Buckmeier. "Oryx was impressed and asked us to create a map of the entire 6,500-square km project area using all 1800 control points."
The result of the second bathymetric mapping project was even better - an average error of one-quarter meter. The difference in the second project was the satellite image. The second image had been acquired on a day when winds were milder and the water was less muddy. The lower sediment content contributed to more accurate mapping.
"We were very pleased with the bathymetric mapping extraction from the satellite data. This information helped in the planning of the seismic survey and will be relied upon heavily for the drilling program," said Reisser.
Applications elsewhere"This process is applicable in any shallow, near-shore environment where the water is relatively clear and recent bathymetric maps are unavailable," said Thetford. He advises shopping around for satellite images before trying the technique.
If several recent archived images are available over a site, Thetford suggests researching historical meteorological information to determine which image was acquired under the lightest wind conditions. Falcon purchased one scene that was completely unusable due to excessive turbidity related to high winds.
"You should know going into the project that you may have to buy several images to get minimal suspended sediment and the best results," he said.
Copyright 1997 Oil & Gas Journal. All Rights Reserved.